Assessing riparian ecosystem water use, particularly transpiration from vegetation and evaporation from soils (‘plant water use’, hereafter), is key to developing sound water management approaches. In western North America, a multidecadal drought is reducing water availability and increasing the use of detailed water budgets. Questions related to both removal of vegetation for water salvage and budgeting water to maintain valuable riparian areas have led to a wealth of studies on riparian plant water use across dryland river systems in North America. Towards evaluating broad patterns in riparian plant water use, we synthesise results from over two decades of research, with the goal of informing water management policies and planning. This study asks: (1) Do some riparian plant communities exhibit lower plant water use than others? (2) Do riparian plant communities have higher water use under hotter climates? (3) Can statistical models based on existing data, plant communities and climate data be used to predict water use for unmeasured locations? Using hierarchical Bayesian models to synthesise data on annual and daily-scale plant water use, we show that marshes, cottonwood-willow stands and tamarisk not impacted by biocontrol use larger amounts of water at the annual scale than other vegetation communities. All plant communities have higher annual water use in hotter climates, which is likely related to a longer growing season and higher evaporative demand. Statistical models based on existing water-use data, plant communities and climate provide bounds on plant water use that can be applied to unmeasured locations and used to evaluate the effects of plant community change on water use. This synthesis produces the most complete summary of riparian plant water use in North American drylands to date and provides water use predictions across different climate and community scenarios that can be used for current and future conditions.